| 13 Jan 2026
Evidence for subglacial flooding in labyrinthine channels on Devon Island, Nunavut, Canada
Abstract. Subglacial drainage systems route glacial meltwater to the ice margin either via efficient, channelized systems or inefficient, distributed systems. The interplay between channelized and distributed drainage systems varies spatially and temporally, governed by meltwater supply and abundance, bed roughness and topography, ice sliding velocity, and ice driving stress. Subglacial channel formation and evolution are therefore affected by variability in meltwater supply to subglacial conduits, and these changes may be recorded in the geomorphology of these channels. The formation of subglacial bedrock channels is attributed to higher energy and/or higher magnitude discharge events, such as the episodic release of meltwater in the form of either subglacial or proglacial floods, in comparison to the energy or discharge required to excavate channels in soft sediment. Common features of landscapes modified by meltwater floods include anastomosing channels and multiple erosive surfaces, wherein the pre-existing drainage system is inundated, resulting in the incision of new channels that reconnect downstream. Devon Island in the Canadian Arctic Archipelago was covered by the thin (<1000 m), cold-to-polythermal based Innuitian Ice Sheet over the course of at least three glacial expansions during the last glacial cycle. Despite this, there is a conspicuous lack of typical glacial landforms, and instead, the inland plateau region of the island is incised by ubiquitous subglacial and lateral meltwater channels. Some sets of bedrock subglacial channels on Devon Island bear a striking resemblance to the morphology of The Labyrinth in Antarctica, which formed by the episodic drainage of a subglacial lake. The characteristics, topology, and morphology of these channels, referred to as 'Labyrinthine channels' hereafter, together with two subglacial channel networks make the focus of this study. We argue that, within both labyrinthine and other subglacial channel networks on Devon Island, the presence of distinct erosional surfaces, anastomosing channels, and profile slope breaks imply formation by short-lived locally intense episodes of erosion. The presence of well-defined erosional surfaces suggests floods progressively incised into lower elevations where meltwater was captured by pre-existing or incipient channels. Moreover, steep contacts between erosional surfaces, termed here as "slope breaks", are similar to fluvial knickpoints and hanging valleys found in other notable landscapes caused by flooding, such as the Channeled Scablands, possibly indicating channel headward erosion in response to pulses of intense erosion. Overall, we suggest that the presence of discrete erosional surfaces implies multiple flooding events, and that changing flow conditions during these events are evidenced by slope breaks. Multiple erosional surfaces, scabland-type landscapes, anastomosing bedrock channels, and hanging valleys with steep slope breaks are not consistent with ice marginal melt, demanding large discharge conditions and pulses of activity, and pointing at subglacial rather than marginal or proglacial environments of formation. This work aids in enhancing the current understanding of the role and dynamics of meltwater drainage systems operating under the cold-to-polythermal based Innuitian Ice Sheet, perhaps shedding light into its retreat dynamics, and bolstering the interpretation of glacial dynamics on Devon Island.
The manuscript by Ruso et al. is an intriguing look at channel networks on Devon Island. The paper begins with a solid review of the literature on subglacial meltwater and channels, folding in proglacial examples with similar morphology to their study site. There is a variety of interesting and detailed measurements of the channels with limited ground truthing. The distinction between ice-marginal vs. subglacial channels could be stronger, and the text could be written more efficiently by deleting much of the commentary. Lastly, there could be more discussion on the longitudinal profiles and whether they make the case for a subglacial origin for the channels. Abstract is quite long and the text is very wordy with too much commentary and use of personal pronouns, in my opinion.
Major comments
Minor comments:
Specific Comments by Line Number.
Ln 13, 14, sentence topic is subglacial bedrock channels that is inconsistent with proglacial channels at end of sentence.
Ln 22, place ‘subglacially’ in front of ‘by’.
Ln 23, unclear what ‘these’ refers to, and remove capital L from ‘labyrinthine’.
Ln 48, Grau Galofre et al. reference year is 2018 in references section, not 2022.
Ln 56, other papers do not argue for time transgressive formation of tunnel channels (Fisher et al. 2022, 2023) in same area as Kehew worked.
Ln 75, hyphenation of cold to polythermal inconsistent within text. Superscript 14 for 14C, and since 100,000 14C years is well beyond limit of radiocarbon dating, perhaps just convert to 100 ka or use marine isotope stage numbers.
Ln 78, provide reference for bedrock once in study area section, and not every time it’s mentioned at the different SG’s. Check capitalization of the rock units.
Ln 87, 88,100, 102: Acronym provided later for IIS, use after first instances, then for every subsequent time.
Ln 97, statement about glacial landforms contradicts line 20 in abstract that there is a conspicuous lack of glacial landforms.
Ln 112, 113, Dyke et al. 2002 and Simon et al. 2015 not included in reference list.
Ln 116, be specific about the glacial erosion features. Are they characteristic of a thawed or frozen bed.
Ln 119, no year for Dyke ref. Who is ‘our’ for ‘our’ work reference? Just the authors on this paper?
Ln 122, no clear what is meant by ‘soil’ is it not fine sediment, or do you mean regolith?
Ln 125, is bedload an indicator between the two types of channels?
Ln 144 missing word, or is ‘as’ to be ‘was’?
Ln 145, Fig. 8B is out of order.
Ln 147, morphology and sedimentology of what? Sort of a repeat of previous sentence.
Ln 154, is the UAV the same one as on line 149?
Ln 157, altitude of 100 m above channel bottom or plateau?
Ln 198, are channels only a meter wide?
Ln 201. Somewhat misleading to use sediment particle sizes for material not transported.
4.2 SG2, it’s interesting that the channels are curvy linear here. Is such a pattern expected subglacially, or more consistent with incision along a retreating ice margin?
Ln 238, what does ‘loosely’ incised mean??
Ln 239, are not the marginal channels incised into bedrock? Sentence needs work as next phrase is confusing.
Ln 244, missing some words after ‘2024’
Ln 248, include more information on the pothole in Fig. 4C. Dimensions etc.
Ln 250, scabland is known to readers from the channeled scabland landscape first described by Bretz. Need to better introduce your use of the scabland word for bedrock slab scabland. Do the slabs have various meltwater sculpted forms on them to give the scab like appearance (grooves, cavettos, sichelwannen, rat tails, etc.)? Features not obvious on Fig 4. If high flow at onset is recorded by the bedrock slab in the bottom of the channel, then channel is pre-existing and little was accomplished by the water.
Ln 279, here you say no channel deposition, but on line 273 you described rounded bedload.
Ln 293, comparison with hanging valleys is fine, but you have knickpoints or hanging channels, not valleys, in the study area, and throughout ms (e.g., ln 453).
Ln 318, don’t see a Fig 6F.
Ln341, ‘funnel-shaped origin’ is not a process but a morphological descriptor.
Ln 375, confusing how drainage can be in two opposite directions.
Ln 428, the lateral meltwater channels are difficult to see in Fig. 10E.
Ln 461, see also Kehew et al., 2009 for example of larger anastomosing channels in sediment / weak bedrock.
Ln 496, 531, terracing rather than layering would be a better description.
Ln 502, 564, cobbles are not large bedload and are easily moved by small streams. Fig 10H does not exist.
Ln 507. Considering that the remnant of the IIS (modern ice caps) are presumably cold based, subglacial floods events are more difficult to imagine especially with thinning ice. How does such an ice cap collapse? Any previous studies on this?
Ln 538-540, confusing sentence.
Ln 556, unclear what ‘their’ refers to. Lakes or channels?
Ln 558, citation not in reference list.
Ln 564, cobbles and incision into bedrock is a weak argument for large discharge. Quantify by what you mean by large discharge. How have discharge calculations been made through the channels? What input data?
Ln566–569, unclear how a cold-based ice sheet over land would suddenly collapse. Moulins and crevasses reopening over same place seems reasonable if driven by subglacial topography. Similar channels are throughout surrounding region so presumably a common process operating.
Figures:
Geographic grids (lat long, UTM) missing from all maps.
Figure 1. black text difficult to read in many places. Use a white line background to make more visible.
Figure 2, use a partially transparent arrow to indicate ice margin recession direction.
Figure 3 and most others, a large range in font sizes. Legend almost unreadable while the text in the symbology box is much too large.
Fig. 4A, white arrows are too faint.
Fig. 12B, the ice surface profile is very misleading.